Selection-free Targeting of the OCT4 Locus using a Mini-vector

To further explore the possibility of making knockin reporter alleles without drug selection, we designed a “mini-vector” donor plasmid, 2A- mOrange, which is similar to 2A-eGFP-PGK-Puro except that there is no PGK-Puro cassette and eGFP was replaced by mOrange (Appendix 7A). We also replaced the crRNA/tracrRNA duplex cr1-dp with the single gRNA cr1 targeting the same sequence (Appendix 7B), as the chimeric version works more efficiently than the original duplex design (Jinek et al., 2012). Similar to the experiment with the 2A-eGFP-PGK-Puro donor, we co- electroporated HUES8 hESCs with a plasmid expressing Cas9/gRNA and the new 2A-mOrange mini-vector (Appendix 7A). In contrast to the absence of fluorescence after integration of the 2A-eGFP-PGK-Puro cassette,

integration of the 2A-mOrange cassette resulted in mOrange expression in

∼0.001% of cells as detected by fluorescence-activated cell sorting (FACS)

(Appendix 7C). One may enrich mOrange-expressing cells for establishing OCT4 reporter lines. However, this low efficiency is impractical for genes not expressed in undifferentiated hESCs, as one has to rely on randomly picking individual colonies to establish clonal lines.

Our recent study shows that the iCRISPR platform enables efficient gene editing using short ssDNA HDR templates (Gonzalez et al., 2014), prompting us to further optimize the iCRISPR platform for HDR using longer circular dsDNA donor vectors. After optimizing transfection conditions, we

co-transfected doxycycline-treated HUES8 iCas9 cells twice in 2 days with the OCT4 cr1 gRNA and the 2A-mOrange mini-vector using Lipofectamine 3000 (Appendix 7D, Appendix 8A-B performed by Zhengrong Zhu). FACS analysis identified ∼0.4% mOrange-expressing cells from the HUES8 iCas9 targeting and ~.24% mOrange-expressing cells from the MEL1 iCas9 targeting (Appendix 7E), supporting the general utility of this new approach in diverse human pluripotent stem cell (hPSC) backgrounds. In contrast electroporation of the HUES8 iCas9 without doxycycline treatment did not show a detectable number of mOrange positive cells (Appendix 7F). The much-improved efficiencies can be partially attributed to the integration of Cas9 in the genome as a ∼5- to 6-fold increase of mOrange+ cells was observed compared to the control condition where iCas9 hESCs (not treated with doxycycline) were transfected with Cas9/gRNA and the donor vector using Lipofectamine 3000 (Appendix 7F). The use of Lipofectamine

transfection also substantially increased the targeting efficiency compared to the electroporation method (Appendix 7E-F). There may be other ways to improve the transfection efficiency (e.g., through nucleofection) to achieve similar results with or without the use of iCas9 hESCs (Byrne et al., 2015). We picked ten colonies from individual FACS-isolated mOrange+ cells and identified six correctly targeted clones by PCR and Southern blot analysis (Appendix 9A-B). All six lines co-expressed mOrange with pluripotency markers such as OCT4, SOX2, and NANOG and displayed normal hESC morphology (Appendix 9C). We further examined the OCT4-mOrange hESC

reporter lines along with the OCT4-eGFP lines for reporter gene expression after differentiation. After 3 days of treatment with BMP4 and SB431542, a TGFβ inhibitor (Hou et al., 2013), hESCs exhibited a differentiated

morphology, and eGFP and mOrange expression were downregulated in the respective OCT4-eGFP and OCT4-mOrange hESC reporter lines with

concomitant loss of endogenous OCT4 expression as determined by immunostaining and FACS analysis (Appendix 10A-B). Thus, the OCT4- eGFP and OCT4-mOrange reporters faithfully reflect endogenous gene expression during the maintenance and differentiation of hESCs.

We next investigated whether the relatively high targeting efficiency was achieved at the expense of undesirable mutations at the OCT4 locus or any off-target sites. All eight OCT4-eGFP and six OCT4-mOrange lines examined showed the expected sequence at the junction between the

endogenous OCT4 sequence and the inserted sequence. This is reassuring, as we made sure that the donor template did not contain the CRISPR target sequence to prevent undesired mutagenesis after reporter gene integration. However, Indel mutations were detected in the non-targeted allele in two of the six OCT4-mOrange reporter lines examined (Appendix 10C). These findings underscore the necessity of thorough sequence analysis for eliminating clones with undesired mutations in the non-targeted allele, a point not widely recognized with the CRISPR/Cas-mediated targeting strategy. We also sequenced seven predicted off-target sites based on the 12-bp seed sequence important for target recognition (Jiang and Pugh,

2009; Jinek et al., 2012). Examination of six OCT4-mOrange and eight OCT4-eGFP lines revealed no mutations except that three OCT4-eGFP lines carried mutations at the POU5F1P4 locus, which shares the same 20- nt target sequence with the intended target (Table 4.1).

Table 4.1 Sequencing analysis of potential off-target sites in OCT4- eGFP and OCT4-mOrange reporter lines

Gene CRIPSR Target Sequence-PAM

Sequencing results in OCT4-eGFP reporter lines

Sequencing results in OCT4-mOrange reporter lines 1 2 4 5 6 7 9 1 0 3 5 6 7 8 1 0 OCT4 TCTCCCATGCATTCAA ACTG-AGG DLG2 AAGCTCAGGCATTCAA ACTG-TGG - - - - - - - GPHN GCCCTCAGGCATTCA AACTG-TGG - - - - - - - IMMP2L TAGACTTAGCATTCAA ACTG-AGG - - - - - - - PEMT GCACCCTAGCATTCAA ACTG-TGG - - - - - - - POLR2J 4 AAGGAGAAGCATTCAA ACTG-TGG - - - - - - - POU5F1 P4 TCTCCCATGCATTCAA ACTG-AGG - HE T* - HOM ** - HOM* ** - - - - - - SLC33A 1 CAGAAATGGCATTCAA ACTG-CGG - - - - - - -

-: Both alleles are wild-type; HET: One allele has a mutation; HOM: Both alleles have mutations;

* 6 bp insertion; ** 4 bp deletion; *** 10 bp deletion.

4.2.3 Selection-free Targeting of the NANOG Locus using a Mini-